How NVIDIA provides AI powered solutions ?

  • Mallika Rangaiah
  • Jul 27, 2021
  • Artificial Intelligence
How NVIDIA provides AI powered solutions ? title banner

Once you’ve acknowledged the power that Artificial Intelligence (AI) holds, there’s no looking back. Technology has revolutionized operations in every field and sector that it has touched. By empowering systems and machines into simulating human cognitive functions on the basis of psychology, neuroscience, and computer science it has revolutionized the operations of each industry from retail, healthcare, finance to media to a considerable extent. 

 

This revolution has just kick-started and it exhibits the scope of infusing each and every aspect of daily human life.  A certain platform that has risen like a phoenix in terms of AI-powered solutions would be the tech giant NVIDIA. While NVIDIA may have come across AI accidentally, there was no looking back once the founders realized the potential this phenomenon held. 

 

Initially starting off as a graphics processing unit producing platform, NVIDIA is now recognized for both its realistic video game graphics as well as for dominating AI chips. 


 

About NVIDIA

 

Founding year: 1993

Headquarters: Santa Clara, California

 

Established by Jen-Hsun Huang, Curtis Priem, and Christopher Malachowsky, NVIDIA was based on the confident belief that one day PC would become a consumer device for enjoying games and multimedia. 

 

In the present times, NVIDIA Corporation is an AI computing company. The platform operates via two segments namely, Graphics and Compute & Networking.

 

It is a global corporation that produces graphics processors, desktop computers as well as mobile technologies. It is recognized for the development of integrated circuits that are applied in devices ranging from personal computers or electronic game consoles. 

 

The platform is among the most powerful global manufacturers of high-end graphics processing units (GPUs). Let’s take a look at how NVIDIA makes use of AI-powered solutions in its diverse operations. 

 

 

How AI is applied in different industries by NVIDIA

 

“Everything that moves will be autonomous someday, whether partially or fully. Breakthroughs in AI have made all kinds of robots possible, and we are working with companies around the world to build these amazing machines.”

- Jensen Huang, CEO, Nvidia 

 

NVIDIA is determined to not leave any stones unturned when it comes to making use of AI in each of its operations. Among the latest updates of the platform, it has also partnered with VMware with the goal of introducing AI to each enterprise and now the platform is all set to unleash the power of AI in the hybrid cloud. 

 

While the platform’s major revenue stream still remains graphics and gaming, yet in the past financial year (2020-2021) its sales of GPUs for use in data centers reached $6.7 billion.

 

In 2020, the platform introduced its new system the NVIDIA A100, which was pegged as “AI supercomputer in a box.” Having over 54 billion transistors, this is one of the most compelling chip systems to be generated. With a single A100 pack holding the same computing power held by 300 data center servers, one A100 conquers an entire room of servers to a single rack. 

 

The platform even offers training to help develop key skills in AI, accelerated data science or accelerated computing for users who desire it through its NVIDIA Deep Learning Institute (DLI). Let’s take a look at how NVIDIA makes use of AI in the different industries it caters to. 

 

 

Media and Entertainment

 

The media and entertainment sector has ceaseless deadlines to meet while being required to supply high-quality content and manage a wide and complex workforce at the same time. This is where NVIDIA’s technologies step in. 

 

Technologies such as AI, simulation, real-time ray tracing, and virtual production allow box office hits, popular TV programs, and legendary pictures to be made possible. These technologies allow for fresh opportunities and the capacity of adopting deep customer insights to generate fresh ways of engagement while arriving at global markets across platforms on each device.

 

 

Retail

 

The solutions offered by NVIDIA allow retailers to minimize shrinkage, enhance forecasting, automate warehouse logistics, decide on strategies for in-store promotions as well real-time pricing and customize the consumer recommendations to offer an improved shopping experience, both offline and online. 

 

For instance, NVIDIA’s support has allowed the tech provider Trigo to revamp grocery stores into massive supercomputers.

 

Trigo has linked AI engines to its cameras and some of its weighted shelves for the small-sized items that might get concealed by the shopper’s hand. Through these sensors, the platform generates a 3D model of the store. The neural networks can detect the items which the consumers insert in their baskets. Once the customers finish shopping and leave, Trigo sends the tally to the store and the store digitally bills the consumer with the amount.


 

Healthcare and Life Science

 

Be it data center, the edge or the cloud, NVIDIA offers solutions streamlining the development and use of domain-specific AI for allowing fresh accomplishments in areas of patient care, research as well as treatment results.

 

An excellent example of NVIDIA in healthcare is that of the imaging platform Paige.AI which has set up an AI system that can transform the process of a cancer diagnosis. The platform has dispensed massive numbers of real-life medical images into its neural network. With the use of 10 NVIDIA GPUs, Paige.AI has trained its system to discover any initial hints of a tumor. 


The image shows how AI is applied in different industries through NVIDIA namely, Media and entertainment industry, Retail, Healthcare and Life science, smart cities, supercomputing and transportation.

How NVIDIA is applying AI in different industries


Smart Cities

 

NVIDIA AI solutions offer compelling new approaches to creating more sustainable cities, maintaining infrastructure, and enhancing public services for the communities and residents.

 

This is done by the platform offering the ability to collect data from a massive group of sensors and other IoT devices and then extracting effective insights via the NVIDIA Metropolis platform for Intelligent Video Analytics (IVA).

 

 

Supercomputing

 

From predicting the weather, discovering drugs to locating fresh sources of energy, NVIDIA GPUs unleash the fastest supercomputers. 

 

Through performance that can reach and surpass petascale performance, supercomputing offers researchers the capacity that they require for simulating and predicting the world. Enhancing productivity and boosting the number of scientific simulations can create a considerable impact on the quality and quantity of scientific advancements. 

 

An example is NVIDIA Jetson TX2, an embedded AI supercomputer that delivers 1 teraflops of performance in a credit card-sized module. This level of capacity allows a fresh wave of manufacturing automation, allows drones to scan hazardous locations, and also enables robots to ship a massive number of packages each day.

 

 

Transportation

 

Through the robust capabilities of its GPUs, NVIDIA ensures that each industry gathers and processes large sections of data, trains models, develops secure, self-navigating vehicles, devices, and robots, and also ensures thorough automation. 

 

Recommended blog - AI in Tesla

 

An example of this would be NVIDIA’s contribution to developing robotaxis. NVIDIA DRIVE offers full-stack AI compute solutions for robotaxi development, allowing fully autonomous vehicles that process massive amounts of data and operate on superfluous and diverse deep neural networks to ensure secure operation.



 

Recent AI-powered solutions offered by NVIDIA Research

 

There’s no end to the AI solutions being presented by NVIDIA. We’ve listed some recent ones developed by NVIDIA Research below : 

 

NVIDIA VID2VID CAMEO

 

Could you imagine having the luxury of just rolling out of bed, switching open your laptop and managing to look fresh and rosy for each video call? Vid2Vid Cameo makes this possible by allowing users to submit a professional, polished, 2D picture or cartoon avatar before a call. The still image can then be adopted as the basis for synthesizing realistic talking head videos by the AI model using GANs. The AI maps the user’s facial movements to secure the real-time motion and altering viewpoints.


 

NVIDIA GAMEGAN

 

The legendary PAC-MAN has been reinterpreted by AI through the AI model GameGAN. Trained on the basis of 50,000 episodes of the game, NVIDIA Research’s GameGAN, is a compelling AI model that can generate a fully functional version of PAC-MAN, without including the basic game engine. 

 

This implies that even without needing to comprehend the fundamental rules of the game, AI is capable of recreating it satisfactorily. This is the first neural network model that simulates the engine of a computer game by exploiting generative adversarial networks (GANs).

 

 

NVIDIA GANIMAL

 

Picturise the smile of a dalmatian being projected on a lion, or a tiger having the expressions held by a bear. While this can be picturized by the human brain, it would be pretty difficult to be executed by a computer. Interestingly, NVIDIA’s GANimal, once it is fed the picture of one animal, holds the power to recreate the expression of that animal and project it onto other animals.  

 

Once an image has been fed upon the GANimal app, its image translation network is capable of projecting the pet’s characteristics upon other animals. 

 

 

NVIDIA GAUGAN

 

This app has been named after the post-Impressionist painter Paul Gauguin and it generates photorealistic images from segmentation maps (labeled sketches depicting a scene’s layout). 

 

GauGAN allows artists to make use of paintbrushes and paint bucket tools for designing their own landscapes through labels such as a cloud, a tree, or a rock. Through a style transfer algorithm, the app enables the artists to apply filters, such as switching a sunset into daytime. It also allows users to upload their own filters to add to their works or make use of custom segmentation maps as well as landscape images to enhance their artwork.

 

 

NVIDIA IMAGE INPAINTING

 

This app makes use of deep learning inferences and empowers them into creating their own personalized portraits or landscape. 

 

Recommended blog - Deep Learning applications

 

It allows the users to fill in the missing portions of an image using new pixels which are developed through the trained model, regardless of what is missing from the image. 

 

 

Enhancing video calls through NVIDIA Maxine

 

NVIDIA Research has introduced an approach of adopting AI for minimizing video call bandwidth while enhancing the quality at the same time. 

 

By substituting the h.264 video codec with a neural network, it can manage to minimize the required bandwidth for a video call by an order of magnitude. Through AI aided video calls, the sender will first submit a reference image of the caller first,  and then rather than delivering a stream of pixel-packed images, it sends specific reference points on the image around prime features i.e the eyes, nose, and mouth.

 

The receiver side then through a generative adversarial network makes use of the reference image along with the provided key points for reconstructing the ensuing images. Since the key points are lesser than the full pixel images, minimal data is sent, so even with a low internet connection, a coherent and clear video chat will be created. 

 

 

NVIDIA Jarvis

 

Released towards the end of February 2021, NVIDIA Jarvis is an application framework for multimodal conversational AI services that delivers real-time performance on GPUs.

 

As per NVIDIA’s website, the Jarvis framework includes pre-trained conversational AI models, tools in the NVIDIA AI Toolkit, and optimized end-to-end services for speech and natural language understanding (NLU) tasks.

 

The Jarvis-based applications have been optimized for enhancing performance on the NVIDIA EGX™ platform in the cloud, in the data center, as well as at the edge.

 

AI is an evolving entity that is a complex and intricate source for developing, expanding, and making use of. Holding more than a decade of experience in developing AI for enterprises worldwide, NVIDIA has established end-to-end AI frameworks and solutions which empower each enterprise to set their AI sparked ambitions ablaze. 

Comments